ArticlesStillbirth rates: delivering estimates in 190 countries
Introduction
Each year 10·6 million children die, yet most child deaths occur uncounted by national health-information systems.1 This dearth of data leaves the world “stumbling around in the dark” on many critical global-health issues.2, 3 Until recently, most attempts to improve information about child deaths focused mainly on the postnatal period. Recognition that almost 40% of deaths in children younger than 5 years occur in the first month of life, and that 30% occur in just the first week, has increased global attention on the period immediately after birth.4 However, stillbirths are yet to be counted in global statistics or indeed in international health policy. Stillbirths are not reported in WHO's routine mortality data or in most population-based surveys; they are not included in the Millennium Development Goals or in estimates of the global burden of disease. However, novel attempts are underway to develop methods to calculate disability adjusted life years for stillbirths.5
There are various epidemiological, programmatic, and rights-based arguments for the measurement of stillbirths. First, counting all births—dead or alive—increases the probability of correctly recording all the important outcomes, including livebirths, stillbirths, and early neonatal deaths. Babies who die very soon after birth are less likely to be registered than are older babies who die, and stillbirths are even less likely to be recorded.6, 7 Additionally, live-born babies who die early might be misclassified as stillbirths and vice versa for several reasons: lack of knowledge; lack of careful assessment for signs of life; avoidance of blame, extra work, or audit review for the birth attendant; or reasons of perceived gain or loss for the family. For example, the registration of a livebirth could encumber the family with funeral arrangements and costs, and the physician with extensive paperwork, whereas a stillbirth requires no funeral and less paperwork—differences that may promote misclassification towards stillbirths. If both stillbirths and early neonatal deaths are counted, then early neonatal deaths misclassified as stillbirths are at least recorded even if they are misclassified.
Second, decision-making for obstetric and neonatal health programmes might be misguided if monitoring does not include stillbirths.8 Data suggest that as obstetric care improves, intrapartum stillbirths might decline but early neonatal mortality could initially rise as babies survive birth but die soon after.9, 10 Hence, if stillbirths are not counted in routine programme monitoring, assessment is incomplete.
Finally, prevention of stillbirths is important. The death of a baby during the last trimester of pregnancy is a source of pain to mothers and fathers, and indeed is reported to be associated with grief reactions more protracted than for early neonatal deaths, partly because of the social taboos associated with open grieving for a stillbirth.11, 12 There are proven interventions used almost universally in rich countries that do not reach enough women in the world's poorest countries; unfortunately, invisibility contributes to inaction.
Historically, stillbirths and early neonatal deaths were grouped together as perinatal deaths. Perinatal epidemiologists are moving away from this practice.13 There are two major arguments for reporting stillbirths and neonatal deaths separately. First, there is much confusion over multiple definitions of perinatal mortality, which cover up to ten time periods depending on the definition of fetal deaths used (20, 22, 23, 24, or 28 weeks of gestation) and how much of the neonatal period is included (from early gestation to day seven, or all the gestation period to day 28).14 There is an absence of comparability even within Europe.15, 16 Advances in neonatal care have pushed the bounds of fetal viability to around 23 weeks in industrialised countries, forcing changes in definitions of fetal death; however, these issues remain largely irrelevant in most countries with high mortality rates. For international comparison, WHO promotes the definition of stillbirth (or late fetal death) as death occurring at at least 28 weeks of gestation or at least 1000 g birthweight.17, 18 We use stillbirth to mean babies born dead during the last trimester of pregnancy. A baby who dies 5 min after birth, or indeed who has a detectable heart rate at birth, counts (at least in principle) in the global estimates of child deaths. A baby who dies even in the process of birth does not count.8
Second, given that under-reporting is a bigger problem for stillbirths than for neonatal deaths,6, 7 combination of the two measures continues to mask the data weakness and to perpetuate the scarcity of quality data available for stillbirth rates. There are few high-quality, population-based data for stillbirths even in industrialised countries,19 and the improved systems for collection of data indicate that stillbirth rates continue to be under-reported.20
WHO produced the first global estimates of perinatal mortality by region for 1983 and again by country for 1995,17 but neither report gave stillbirth rates at national or regional levels. For the first time, the World Health Report, 2005, included stillbirth rates for some countries and gave a global total of 3·3 million.21 Sources and methods are not yet available.
Global statistics groups and WHO are actively promoting a systematic and transparent approach to global estimates with “well-documented, preferably peer-reviewed, and published methods of estimation”.22 Four steps are recommended—accessible databases, transparent methods, an independent advisory group, and overall consistency through clearance procedures.23
We used vital registration, survey, and study-based data that met preset inclusion criteria to estimate stillbirth rates and the numbers of stillbirths with uncertainty estimates for 190 countries, for the year 2000, providing an accessible list of data inputs and applying transparent and replicable methods.
Section snippets
Data
We used data for stillbirth rates from three sources (figure 1). First, we used available vital registration data from 32 developed and 12 developing countries. We included only countries that reported 90% or greater completeness of recording of adult deaths.24 These countries consist of those reporting stillbirth data to the Council of Europe25 and those responding to our direct requests for data from vital registration offices (webtable). Second, we used demographic and health surveys (DHS)
Results
The final dataset consisted of 323 observations from 103 countries with data-collection years ranging from 1976 to 2003. All world regions were represented (figure 2). The mean reference year from the study observations was 1995 for developed countries and 4–7 years earlier for all developing country regions, apart from Eurasia for which the mean reference year was 1999. All the vital registration-based estimates were for the years 1999–2001, apart from Denmark (1996) and Bosnia Herzegovina
Discussion
We have provided systematic global estimates for stillbirths at country level, detailing inputs and methods and providing uncertainty estimates. The global total of 3·2 million stillbirths is similar to the World Health Report 2005 (3·3 million),21 but is less than previous WHO global estimates of 4 million in 1999,29 and 5·3 million in 1995.17 However, our point estimate of 3·2 million has wide uncertainty estimates with a range of 2·5–4·1 million.
The basic difficulty faced in estimating
References (37)
- et al.
WHO estimates of the causes of death in children
Lancet
(2005) - et al.
Gaps in policy-relevant information on burden of disease in children: a systematic review
Lancet
(2005) Stumbling around in the dark
Lancet
(2005)- et al.
4 million neonatal deaths: When? Where? Why?
Lancet
(2005) - et al.
Long-term impact of perinatal bereavement: comparison of grief reactions after intrauterine versus neonatal death
Eur J Obstet Gynecol Reprod Biol
(1997) - et al.
The availability of perinatal health indicators in Europe
Eur J Obstet Gynecol Reprod Biol
(2003) - et al.
Better health statistics are possible
Lancet
(2006) The Ellison Institute: monitoring health, challenging WHO
Lancet
(2005)- et al.
Incorporating deaths near the time of births into estimates of the global burden of disease
- et al.
Why are Thai official perinatal and infant mortality rates so low?
Int J Epidemiol
(1990)